Agenda
Chairs: Mario Giraldo
Introduction
Mario Giraldo
Bayes theorem and concepts for its application into clinical diagnosis for rehabilitation
Mario Giraldo
Practical exercises to estimate the probabilities of having a disease from already published data
Attendees
Discussion: analysis of the exercises and the implications for everyday regular clinical practice, research or medical education
Mario Giraldo
Closing
Mario Giraldo
Session outline
Introduction: A diagnostic test may not include the false positive or negative rates, so it leads the clinician to a yes/no statement of a diagnosis with missing critical information. Consequently, the information to the patient and therapeutic decisions could be misleading in non-typical cases where the diagnostic test does not have the accuracy to confirm or to rule out the diagnosis. Clinicians may easily use and alternative statistical method based on the Bayes Theorem. Development of the session: This is a mixed theoretical-practical session to calculate the probabilities to diagnose a pathology by using the Bayes Theorem. A theoretical background compares the frequentist statistics versus the Bayesian statistics on how to get closer to the “truth” as a mainstay for the medical and rehabilitative scientific literature. The attendees will be trained to calculate the Bayesian probabilities of a diagnosis either positive or negatives results given actual cases of shoulder tendinopathy, neurogenic or arterial claudication, neurogenic bladder, or carpal tunnel syndrome. The attendees will learn to overcome the false positive or negative results given the calculated probabilities. These calculations may help the clinician to think of reasonably confident ideas, when the probability is the lowest or the highest, and therefore, to believe that the pathology is absent of present, respectively. On the other hand, the clinician may request a new test to improve the certainty when the estimated probability does not point towards the absence neither the presence of the disease, due to the particularities of the patient or the weak accuracy of the test. The understanding of the Bayesian probabilities provides new skills for individualized medicine. (1-5)
- Murthi AM, et al. J Shoulder Elbow Surg. 2000;9: 382-5.
- Cardoso A,et al_J Shoulder Elbow Surg2019;28:2272-78
- Chen H, et al. Ultrasound Med Biol 2011; 37: 1392-8
- Fagan T. NEJM 1975: 293-57
- Farooqi AS, et al. Orthop J Sports Med. 2021 Oct 11;9(10)PMID: 34660823
Learning outcomes
The theoretical-practical workshop will enable the attendees to reach the following goals:
- To recognize the difference between the frequentist statistics and the Bayesian statistics at a clinician level of understanding.
- To understand the limitations of the frequentist statistics in terms of its lack of information about the false positives or false negatives in a particular patient.
- To be able to understand the main components of the Bayesian theorem that are required to calculate a probability to rule out or to confirm a diagnosis in a particular patient.
- To be able to calculate a probability to diagnose a pathology in a particular patient by using the Bayes theorem at a clinician level of understanding.
- To practice a decision-making process based on the estimated probabilities to decide whether a new test is required, or the diagnosis was already established.
Target audience
- Medical practitioners
- Students
- Trainees